The name Kafka now gets invoked whenever someone doesn't immediately get what he/she wants. Some administrative thingy gone wrong? Kafka! Your broadband connection doesn't allow you to download at 20Mb and the help desk says that the speed is not constant? Kafka! Your microwave's remote control's batteries are not in stock at your local supermarket and it will take more than an hour to restock? Kafka! You wake up and you find yourself turning into a giant beetle? O wait...
Re:ah yes, anti-perl tirades are refreshing
on
Coders At Work
·
· Score: 1
That's sort of what I implied. However, I've seen people try to write parser generators in object oriented Perl and that's taking it too far, let alone a real piece of software. I remember some problems I had with Perl. I overcame them and then Perl was upgraded...
Visual programming is definitely not the answer to all problems, but it can help build simple variations on a theme. I can imaging that building simple database driven websites could be done with visual programming. Certain simulations can also done by visual programming, if you've got the right environment. Did you ever look into analog computing? That's a great example of visual programming (and totally esoteric!).
Any complex visual programming will introduce bugs. Different bugs than the ones we know, but bugs nonetheless. If you connect 20 modules, each of which a consists of many modules consisting of many operations, chances are that you've overlooked one or two assumptions and your application will loop indefinitely, produce the wrong answer, whatever. Then try to find the error...
Your point about parallelism is not completely correct. In a purely functional language, everything can be parallelized, even addition, since there are no side effects to worry about. That said, even the lowliest operation has to wait until its operands are available, so you still have a proper sequence of execution. This, in combination with lazy evaluation, was very promising... 25 years ago. Still, if you've written a functional application, you can parallelize it at any point, which is cool.
Re:ah yes, anti-perl tirades are refreshing
on
Coders At Work
·
· Score: 1
If that purpose is making life difficult for programmers in general and obfuscation in particular, then I agree. If the purpose is doing something with/etc/passwd that you can't do with grep, I'm already less inclined to agree. Unless of course the word masterpiece was used in the sense that the atomic bomb is a masterpiece of physics and engineering, then I fully underwrite your sentiments.
If gender is a social construct, then 1. To know if someone is female or male, you would have to ask other people (since they are the labelers), or 2. The difference is so unclear that it is necessary to abolish the segregation between male and female contests. The consequences of both can be rather unpleasant.
I don't know what your point of view is, but gender might be no more a social construct than an elephant is. We, language users and world observers, have classified nature, and gender is one of those classifications. This one happens to be based on reproductive capacities. Now either you conclude that those capacities are also a social construct or gender has a basis in reality.
If you go for the first option, then we can call anyone an elephant, or a horse, and request that they be excluded from races since the races are for people and people alone. If you go for the second option, there must be some consensus on biological criteria for gender. That they are not always recognized on the outside is a different matter.
He spreads the myth that religion and evolution can't be compatible (why exactly could it not be argued that god designed life with the capability to mutate?)
I just finished reading the book, and I think his point is that the God idea of the major religions is incompatible with evolution theory or science in general. He cannot and does not argue against pantheism and other vague notions of there-must-be-something-more-than-this. And he argues quite straightforward that God, the almighty creator, is improbable and that religions are based on an enormous amount of nonsense. And that's no myth.
His hard line approach makes Atheists as a whole look like intolerant arses and I don't want to be associated with it.
No, just him, but he clearly says that he wants religion out of the world, since it's going to kill us. We have developed the power to obliterate the planet before we developed a rational way of dealing with it, to paraphrase Maher.
Yes science changes over time but so does religion.
Panta rhei, but that doesn't make religion one bit better. Anyway, I haven't seen much change in the major religions' holy books over the last millenium.
I don't consider sections saying "Because media character demographics and portrayals of social groups may influence players' likelihood of attending to and learning from game characters" or "Many have suggested that games function as crucial gatekeepers to interest in technology" very strong. And this paper doesn't test such theories, it just takes them for granted.
My point 2 doesn't follow, it's an example: if games can be proven to make their players aggressive, that would demonstrate a clear causal relationship between player and game, right? Tons of studies have tried to prove it, none have been convincing enough to ban such games, and others found no such relation. Such change in behavior coming from games would be rather basic, more basic than a change in willingness to participate in building STEM career improving skills. If the former can't be found, how can you possibly find something so indirect? How can you possibly prove a causal relationship between representation and that change in willingness? This paper doesn't prove anything, except that women (etc.) are underrepresented as characters in the top-selling games (a methodological fallacy that's worth another look, because who buys these games?, but ok).
Ad 3: I think the key is education and awareness/consciousness. Start teaching basic physics in kindergarten (ok, that's too early, I know my Piaget). Force them into useful studies, because it is clear that one of the basic assumptions is that women and minorities should pursue STEM careers. Don't let them get off the hook if they fail their first math test, but give them remedial teaching.
Forget about games. Women don't like violent games as much as men do, and that's not because there is a male lead character, but because they don't like the running, the mazes, the shooting. There were games in which you could choose your role (Unreal, I think). Hell!, the most famousest of famous games had a female character: Lara Croft. Did women like to play Tomb Raider that much? Or was it rather 13 yr old Hispanic boys?
Anyway, the experiment is not hard to set up. Ask EA to produce two different versions of a specific game, one with predominantly male characters, the other with predominantly female characters. Now do a 2x2 design (player gender, game gender), give these people the game and test how well they liked it after one year or so, how often they played it, etc., and see if there is any correlation between that and their results on math exams or their choice of subsequent career.
Ad 4. You have a point. I tried to say that the "secondary ideal" of having women and minorities take up STEM careers was a subjective/political choice, and got carried away in the possible negative consequences that would go against higher ideals (basic equality, freedom, minimum levels of well-being for everyone, etc.) from which it had been derived. And in fact, I do wish more women would go and study physics and maths, but the sad fact is that students have been brought up thinking "who wants to be a physicist if you can be his (or her?) boss". I should have concluded that this again shows that education is the key, and that society has a problem translating its values into reward. If you can make more money being a fashion reporter for Marie-Clair than working at the LHC, there is something wrong, and it's not gender distribution in games.
Ad 1. How does participating in the forms you describe build any skills that is exclusive to the games from the study? Women participate heavily in e.g. The Sims forums, so they do get the same opportunities as everybody else.
Ad 2. There is also enough reason to doubt the outcome of these studies. If they could be taken for true, violent games should be banned immediately.
Ad 3. You're going into a regression there. Why didn't women participate in chess clubs? And is there a relation to interest in science? I can tell you from personal experience that female participation in science is highly culture dependent. E.g., in the former Soviet satellites, the percentage of women in hard sciences was much higher than it is today. In many western countries, the participation of men in exact sciences is much lower than it was 20 to 30 years ago. In contrast, gaming participation has increased from almost nothing to the level of obsession and addiction.
Ad 4a. That just means that education fails or that some other cultural influence makes these groups avoid STEM. Ad 4b. There will always be poor groups. It is unavoidable. When one group gets richer, another gets poorer. If you were to make all poor women and minorities take up math and science, either the value of engineers and scientists would drop (causing a pig cycle and probably deteriorating scientific research), or another group would go socially downwards, which would be tit-for-tat, and cause the need for special programs for this group to gain status, etc.
It's all word play. Yes, there is scientific fraud, conscious and unconscious, and changing equipment to give the output you want falls in that broad category. Of course, our measurements are only as reliable as the equipment we use and the assumptions you rely on (nice example is the history of the speed of light). At least under certain circumstances, it is possible to eliminate the influence of bad equipment, by counter-balancing the use of equipment in the design of the experiment.
Interpretation of the output of non-instrumental computer models is different. A computer model can be seen as a specific instance of a theory and as such should be modified to fit the facts.
Latour's idea about nature is mystical. Nature is the ultimate subject of study. In a normal experiment, we assume that nature simply does what it always does, and that's good enough. If you repeat an experiment, nature will respond the same all over again (modulo fundamental uncertainty). If you don't accept that, you shouldn't be in science. If you do accept it, the conclusion is that nature does put boundaries on our observations, and hence on the course of science, even though people can keep up their belief in disproven theories (e.g. astrology) for quite a long time.
In another thread I wrote this about it: "Have you got any idea how difficult it is to refute an experimental outcome, at least in the less exact sciences? It's not only that you can create a gazillion possible deviations between your set-up and the one from the article (making direct comparison difficult), you will also need to run it with a pretty large subject group if you want to have enough power (making it expensive and time consuming), and then you're going to have problems publishing your article (reviewers and editors don't like null effects). In short, there is no profit in it. Most people, and researchers are people, are in it for the money, prestige, whatever, and replicating a study generally doesn't get you funding, prestige, publications. So guess what happens? The world, at least the part that does experimental psychology, gets stuck with 90% junk publications. And that's being conservative."
I concur with your idea about data sharing. I had thought of it before, but it will only work if researchers share *all* their data, not just the data on which their publications are based. And this data should be accompanied by a standardized description of the experimental situation, stimuli, and subjects. Somehow I think this is not going to happen...
Regrettably, they know that, but they don't care. It's probably some sort of cognitive dissonance reduction: yeah, it took me so many experiments that this outcome can be attributed to chance, but I think I'm on to something and it fits so nicely with my predictions! And everybody else does it, so why can't I?
I did start, a long time ago, but I couldn't stand his rhetoric. Scientists are people, so they behave like people. Big deal. But he just wants you to get the idea that he attacks the gods of modern time by suggesting that scientific facts aren't all they're cracked up to be, while at the same time doing nothing of the kind and pretending to be a scientist himself. The first part goes down very well with the post-modern literary folk, but it doesn't contradict scientific outcome itself. Sure, the process has its errors, but, as I said, scientists are only human. The course that the search takes is steered by society, but it is nature that puts bounds on its path. Unfortunately, the bounds are softer in social sciences than in physics. Anyway, Latour never proved any of this hypotheses, so what the heck are we talking about. It's just fabricated anecdotes.
That's even worse. They would first have to prove: 1) women do not "participate" (what ridiculous word is that in the context of gaming anyway, let's call it play), because of the absence of other representations, and 2a) that games in which other representations are absent promote building such skills, and 2b) that games in which they are present do not build such skills, and 3) that such skills are useful in persuing "STEM" careers, and 4) that it is generally desirable that women do so. And that is overlooking the definition of the skill set.
Ad 1: women don't like FPS and stuff like that, so those games don't matter in the equation. Women like The Sims, where, lo and behold, representation is player determined. Ad 2: there is some evidence that certain types of games help develop peripheral vision, but none that male-dominated games build the any other skills, nor that games like The Sims don't build them. Ad 3: I followed a STEM career (all 4 letters apply to me) and I started studying when Pong was the most widely available game. I don't know where you got the idea that video gaming somehow turns women into scientists. Ad 4: Highly subjective, if not political.
And anyway, youth has never been discouraged from gaming. I think there are quite some hard-core gamers in the younger population, possibly *because* they are underrepresented. Which 14 year old wants to play a 14 year old?
I know it isn't limited to psychology, and the state of affairs you recount looks very much like the one in social sciences. Partially funding is to blame, partially the editorial process of most journals. In my field, it is hard to get a theoretical argument published, and even harder to publish a null result. I actually once got a rejection of a simple logical refutation of some article's conclusion using the data published in that article on the grounds that it didn't contain new data.
Funding depends on publications and publications depend on positive results, which skews the whole scientific search towards trivial variations on existing experiments with the sole purpose of supporting one's own pet theory. That brings power to the accountants of science, not to the thinkers or the unorthodox experimenters. I know that not much good comes from trusting someone because he or she is a "deep thinker" or relying on unorthodox methods, but the current process excludes them beforehand, resulting in a limitation rather than expansion of our knowledge.
I've seen people get director of a Max Planck Institute who got their fame out of experiments that were obviously overinterpreted, only because it fitted the craze of the period (neuro stuff in the 90's). They then went on to raise researchers in their image and guess what: these people now promote ideas that are just small modifications on the ones of their mentor/mentrix as the next big thing, and nobody never ever took the time to do a few simple experiments to counter the criticism against what is perceived by many to be a meaningless interpretation of data.
Anyway, enough rambling for today. Let's hope that not all proper work is laid to waste and that time will separate the wheat from the chaff.
Well, after working in psycholinguistics for 20 years (of which the last 4 in "cognitive neuropsychology"), I have come to the conclusion that not even all data can be trusted, let alone the conclusions. There are many cases of experimenters running dozens of experiments with slightly different conditions before hitting on one that gives them the desired p 0.05. In fMRI experiments, researchers often take more than a year to analyze the results over and over again, changing between different methods of analysis (ROI, threshold, smearing, boxing, statistic, etc.) and then publish the one they (or their supervisor) like best. Or making SPSS return all cross-correlations on questionnaires with over 200 questions and then drawing conclusions from the set of the most significant ones.
So yes, quite a lot of research from "neuro and social psychology" is worthless. I'd say about 90% of it. The problem is just finding out which 10% is valid. And don't start me on replication. Open any journal on experimental psychology and show me articles that replicate a previous experiment *exactly*.
It's not the method that causes the problem, it's the hypothesis. There is some weird, unwarranted reasoning in going from "in a perfect society, gender, race, religion, etc. don't matter" to a null hypothesis that says that the number of male and female characters in a video game should be equal, or any other assumption about distributions. Firsty, the original starting point is not hard (as in gravity or DNA), but subjective. Secondly, the connection between the real world and a video game is superficial at best.
We can also discuss methods, of course, if you feel more comfortable (think representative sample, not being able to manipulate the variables of interest, etc.), because one thing I've learned in my 20 year career in the social sciences is that method is often considered more important than proper reasoning and experimental set-up, usually because that tends to get in the way of obtaining easily publishable results.
You read the numbers in a weird way. The pages of Secunia say:
- Secunia has issued a total of 193 Secunia advisories in 2003-2009 for Microsoft Windows Server 2003 Enterprise Edition. Currently, 6% (12 out of 193) are marked as unpatched with the most severe being rated Less critical - Secunia has issued a total of 130 Secunia advisories in 2003-2009 for Apple Macintosh OS X. Currently, 4% (5 out of 130) are marked as unpatched with the most severe being rated Moderately critical
At least on OSX, the most critical vulnerability requires you to download and mount a malcrafted disk image, which most likely can only crash your system, and perhaps expose kernel memory.
Furthermore, it is 4 against 12 unpatched (in favor of OSX), and of course the other Windows systems (XP, Vista, 7) have got a different set of vulnerabilities. And your language and mark-up suggest mild paranoia. You have not been attacked, and the Pravda doesn't deal with Windows viruses. Get treatment, or move to Montana and join the militia.
It's not going to happen in ten years, and I sort of know what I'm talking about (I've been working on neural models of language processing for quite some time now). First, the brain is huge, really huge. The number of "connections" isn't even known, let alone how everything interconnects. Second, even at the level of a single neuron, there are quite a few problems to be solved. If we could figure out the exact structure of a human brain, but emulate it just using integrate-and-fire type neurons, I predict it would fail miserably. Third, brains change. We don't know how we start out, but our brain changes from learning (if you want to be bewildered: development of cat vision is fascinating). For our brain to reach a stage in which it could be subject to a mental disorder takes some 15 years of normal life, i.e. inside a body that functions in a society. So even if someone could build a baby brain in 10 years, and incorporate it into a very human looking body, it would still take 15 years to develop and then might just not have developed a mental disorder.
Fourth, we don't have a clue as to where to look for mental disorders. Of course, if you have a virtual brain, you could tweak parameters, but not only are these parameters at this moment unknown, it would also be unethical (for it would hurt a person, see also other threads), and still this would not offer more insight into the emergent property that mental disorder is as comparing people with these disorders at this moment.
I don't like Bonferroni at all, but it does give an indication that the outcome is hardly as significant as the author would like us to believe. But the method of analysis is wrong to begin with. Shouldn't it have been either Kruskall-Wallis (and then use all levels instead of just Excellent vs. non-Excellent) or a chi square? And if he would have included "Very good" at maths, the outcome would have been very different.
So we're just looking at post-hoc interpretation of bad statistics here. Move along, please...
That, and the fact that he fails to publish his method of analysis. What was his hypothesis exactly? He didn't seem to have one in advance, so it seems rather post-hoc. And I don't see where he got his p.01 from, because my intuition tells me 44% vs 56% from a group of 27 respondents cannot be that significant (binomial test). He should have tested the interaction, but the page he refers to has no way to test that, so I think we can take the results with the well-known grain of salt.
We know the bit about lies, damned lies and statistics, of course, but bad statistics...
You are wrong. It is very likely that the user group of this "Mechanical Turk" is highly biased. You *could* say that the conclusion is valid for people belong to this group with a propensity to answer questions of the kind "is a photoshopped nude image of Miley Whatever-her-last-name-is illegal".
As to the discovery of "default", the word exists since the 13th century, according to Merriam-Webster:
Etymology: Middle English defaute, defaulte, from Anglo-French, from defaillir to be lacking, fail, from de- + faillir to fail Date: 13th century
and the expression "in default of" means "in the absence of". The word "default" was simply chosen by computer scientists to describe what happens in case of absence of specification.
But then again, the whole 13th century probably took place somewhere in the 60's.
Slashdot Mirror
PONG didn't have a story line either, and what's good enough for PONG is good enough for me!
The name Kafka now gets invoked whenever someone doesn't immediately get what he/she wants. Some administrative thingy gone wrong? Kafka! Your broadband connection doesn't allow you to download at 20Mb and the help desk says that the speed is not constant? Kafka! Your microwave's remote control's batteries are not in stock at your local supermarket and it will take more than an hour to restock? Kafka! You wake up and you find yourself turning into a giant beetle? O wait...
That's sort of what I implied. However, I've seen people try to write parser generators in object oriented Perl and that's taking it too far, let alone a real piece of software. I remember some problems I had with Perl. I overcame them and then Perl was upgraded...
Visual programming is definitely not the answer to all problems, but it can help build simple variations on a theme. I can imaging that building simple database driven websites could be done with visual programming. Certain simulations can also done by visual programming, if you've got the right environment. Did you ever look into analog computing? That's a great example of visual programming (and totally esoteric!).
Any complex visual programming will introduce bugs. Different bugs than the ones we know, but bugs nonetheless. If you connect 20 modules, each of which a consists of many modules consisting of many operations, chances are that you've overlooked one or two assumptions and your application will loop indefinitely, produce the wrong answer, whatever. Then try to find the error...
Your point about parallelism is not completely correct. In a purely functional language, everything can be parallelized, even addition, since there are no side effects to worry about. That said, even the lowliest operation has to wait until its operands are available, so you still have a proper sequence of execution. This, in combination with lazy evaluation, was very promising... 25 years ago. Still, if you've written a functional application, you can parallelize it at any point, which is cool.
If that purpose is making life difficult for programmers in general and obfuscation in particular, then I agree. If the purpose is doing something with /etc/passwd that you can't do with grep, I'm already less inclined to agree. Unless of course the word masterpiece was used in the sense that the atomic bomb is a masterpiece of physics and engineering, then I fully underwrite your sentiments.
If gender is a social construct, then
1. To know if someone is female or male, you would have to ask other people (since they are the labelers), or
2. The difference is so unclear that it is necessary to abolish the segregation between male and female contests.
The consequences of both can be rather unpleasant.
I don't know what your point of view is, but gender might be no more a social construct than an elephant is. We, language users and world observers, have classified nature, and gender is one of those classifications. This one happens to be based on reproductive capacities. Now either you conclude that those capacities are also a social construct or gender has a basis in reality.
If you go for the first option, then we can call anyone an elephant, or a horse, and request that they be excluded from races since the races are for people and people alone. If you go for the second option, there must be some consensus on biological criteria for gender. That they are not always recognized on the outside is a different matter.
He spreads the myth that religion and evolution can't be compatible (why exactly could it not be argued that god designed life with the capability to mutate?)
I just finished reading the book, and I think his point is that the God idea of the major religions is incompatible with evolution theory or science in general. He cannot and does not argue against pantheism and other vague notions of there-must-be-something-more-than-this. And he argues quite straightforward that God, the almighty creator, is improbable and that religions are based on an enormous amount of nonsense. And that's no myth.
His hard line approach makes Atheists as a whole look like intolerant arses and I don't want to be associated with it.
No, just him, but he clearly says that he wants religion out of the world, since it's going to kill us. We have developed the power to obliterate the planet before we developed a rational way of dealing with it, to paraphrase Maher.
Yes science changes over time but so does religion.
Panta rhei, but that doesn't make religion one bit better. Anyway, I haven't seen much change in the major religions' holy books over the last millenium.
I don't consider sections saying "Because media character demographics and portrayals of social groups may influence players' likelihood of attending to and learning from game characters" or "Many have suggested that games function as crucial gatekeepers to interest in technology" very strong. And this paper doesn't test such theories, it just takes them for granted.
My point 2 doesn't follow, it's an example: if games can be proven to make their players aggressive, that would demonstrate a clear causal relationship between player and game, right? Tons of studies have tried to prove it, none have been convincing enough to ban such games, and others found no such relation. Such change in behavior coming from games would be rather basic, more basic than a change in willingness to participate in building STEM career improving skills. If the former can't be found, how can you possibly find something so indirect? How can you possibly prove a causal relationship between representation and that change in willingness? This paper doesn't prove anything, except that women (etc.) are underrepresented as characters in the top-selling games (a methodological fallacy that's worth another look, because who buys these games?, but ok).
Ad 3: I think the key is education and awareness/consciousness. Start teaching basic physics in kindergarten (ok, that's too early, I know my Piaget). Force them into useful studies, because it is clear that one of the basic assumptions is that women and minorities should pursue STEM careers. Don't let them get off the hook if they fail their first math test, but give them remedial teaching.
Forget about games. Women don't like violent games as much as men do, and that's not because there is a male lead character, but because they don't like the running, the mazes, the shooting. There were games in which you could choose your role (Unreal, I think). Hell!, the most famousest of famous games had a female character: Lara Croft. Did women like to play Tomb Raider that much? Or was it rather 13 yr old Hispanic boys?
Anyway, the experiment is not hard to set up. Ask EA to produce two different versions of a specific game, one with predominantly male characters, the other with predominantly female characters. Now do a 2x2 design (player gender, game gender), give these people the game and test how well they liked it after one year or so, how often they played it, etc., and see if there is any correlation between that and their results on math exams or their choice of subsequent career.
Ad 4. You have a point. I tried to say that the "secondary ideal" of having women and minorities take up STEM careers was a subjective/political choice, and got carried away in the possible negative consequences that would go against higher ideals (basic equality, freedom, minimum levels of well-being for everyone, etc.) from which it had been derived. And in fact, I do wish more women would go and study physics and maths, but the sad fact is that students have been brought up thinking "who wants to be a physicist if you can be his (or her?) boss". I should have concluded that this again shows that education is the key, and that society has a problem translating its values into reward. If you can make more money being a fashion reporter for Marie-Clair than working at the LHC, there is something wrong, and it's not gender distribution in games.
Ad 1. How does participating in the forms you describe build any skills that is exclusive to the games from the study? Women participate heavily in e.g. The Sims forums, so they do get the same opportunities as everybody else.
Ad 2. There is also enough reason to doubt the outcome of these studies. If they could be taken for true, violent games should be banned immediately.
Ad 3. You're going into a regression there. Why didn't women participate in chess clubs? And is there a relation to interest in science? I can tell you from personal experience that female participation in science is highly culture dependent. E.g., in the former Soviet satellites, the percentage of women in hard sciences was much higher than it is today. In many western countries, the participation of men in exact sciences is much lower than it was 20 to 30 years ago. In contrast, gaming participation has increased from almost nothing to the level of obsession and addiction.
Ad 4a. That just means that education fails or that some other cultural influence makes these groups avoid STEM.
Ad 4b. There will always be poor groups. It is unavoidable. When one group gets richer, another gets poorer. If you were to make all poor women and minorities take up math and science, either the value of engineers and scientists would drop (causing a pig cycle and probably deteriorating scientific research), or another group would go socially downwards, which would be tit-for-tat, and cause the need for special programs for this group to gain status, etc.
It's all word play. Yes, there is scientific fraud, conscious and unconscious, and changing equipment to give the output you want falls in that broad category. Of course, our measurements are only as reliable as the equipment we use and the assumptions you rely on (nice example is the history of the speed of light). At least under certain circumstances, it is possible to eliminate the influence of bad equipment, by counter-balancing the use of equipment in the design of the experiment.
Interpretation of the output of non-instrumental computer models is different. A computer model can be seen as a specific instance of a theory and as such should be modified to fit the facts.
Latour's idea about nature is mystical. Nature is the ultimate subject of study. In a normal experiment, we assume that nature simply does what it always does, and that's good enough. If you repeat an experiment, nature will respond the same all over again (modulo fundamental uncertainty). If you don't accept that, you shouldn't be in science. If you do accept it, the conclusion is that nature does put boundaries on our observations, and hence on the course of science, even though people can keep up their belief in disproven theories (e.g. astrology) for quite a long time.
In another thread I wrote this about it: "Have you got any idea how difficult it is to refute an experimental outcome, at least in the less exact sciences? It's not only that you can create a gazillion possible deviations between your set-up and the one from the article (making direct comparison difficult), you will also need to run it with a pretty large subject group if you want to have enough power (making it expensive and time consuming), and then you're going to have problems publishing your article (reviewers and editors don't like null effects). In short, there is no profit in it. Most people, and researchers are people, are in it for the money, prestige, whatever, and replicating a study generally doesn't get you funding, prestige, publications. So guess what happens? The world, at least the part that does experimental psychology, gets stuck with 90% junk publications. And that's being conservative."
I concur with your idea about data sharing. I had thought of it before, but it will only work if researchers share *all* their data, not just the data on which their publications are based. And this data should be accompanied by a standardized description of the experimental situation, stimuli, and subjects. Somehow I think this is not going to happen...
Regrettably, they know that, but they don't care. It's probably some sort of cognitive dissonance reduction: yeah, it took me so many experiments that this outcome can be attributed to chance, but I think I'm on to something and it fits so nicely with my predictions! And everybody else does it, so why can't I?
I did start, a long time ago, but I couldn't stand his rhetoric. Scientists are people, so they behave like people. Big deal. But he just wants you to get the idea that he attacks the gods of modern time by suggesting that scientific facts aren't all they're cracked up to be, while at the same time doing nothing of the kind and pretending to be a scientist himself. The first part goes down very well with the post-modern literary folk, but it doesn't contradict scientific outcome itself. Sure, the process has its errors, but, as I said, scientists are only human. The course that the search takes is steered by society, but it is nature that puts bounds on its path. Unfortunately, the bounds are softer in social sciences than in physics. Anyway, Latour never proved any of this hypotheses, so what the heck are we talking about. It's just fabricated anecdotes.
That's even worse. They would first have to prove: 1) women do not "participate" (what ridiculous word is that in the context of gaming anyway, let's call it play), because of the absence of other representations, and 2a) that games in which other representations are absent promote building such skills, and 2b) that games in which they are present do not build such skills, and 3) that such skills are useful in persuing "STEM" careers, and 4) that it is generally desirable that women do so. And that is overlooking the definition of the skill set.
Ad 1: women don't like FPS and stuff like that, so those games don't matter in the equation. Women like The Sims, where, lo and behold, representation is player determined.
Ad 2: there is some evidence that certain types of games help develop peripheral vision, but none that male-dominated games build the any other skills, nor that games like The Sims don't build them.
Ad 3: I followed a STEM career (all 4 letters apply to me) and I started studying when Pong was the most widely available game. I don't know where you got the idea that video gaming somehow turns women into scientists.
Ad 4: Highly subjective, if not political.
And anyway, youth has never been discouraged from gaming. I think there are quite some hard-core gamers in the younger population, possibly *because* they are underrepresented. Which 14 year old wants to play a 14 year old?
I know it isn't limited to psychology, and the state of affairs you recount looks very much like the one in social sciences. Partially funding is to blame, partially the editorial process of most journals. In my field, it is hard to get a theoretical argument published, and even harder to publish a null result. I actually once got a rejection of a simple logical refutation of some article's conclusion using the data published in that article on the grounds that it didn't contain new data.
Funding depends on publications and publications depend on positive results, which skews the whole scientific search towards trivial variations on existing experiments with the sole purpose of supporting one's own pet theory. That brings power to the accountants of science, not to the thinkers or the unorthodox experimenters. I know that not much good comes from trusting someone because he or she is a "deep thinker" or relying on unorthodox methods, but the current process excludes them beforehand, resulting in a limitation rather than expansion of our knowledge.
I've seen people get director of a Max Planck Institute who got their fame out of experiments that were obviously overinterpreted, only because it fitted the craze of the period (neuro stuff in the 90's). They then went on to raise researchers in their image and guess what: these people now promote ideas that are just small modifications on the ones of their mentor/mentrix as the next big thing, and nobody never ever took the time to do a few simple experiments to counter the criticism against what is perceived by many to be a meaningless interpretation of data.
Anyway, enough rambling for today. Let's hope that not all proper work is laid to waste and that time will separate the wheat from the chaff.
Well, after working in psycholinguistics for 20 years (of which the last 4 in "cognitive neuropsychology"), I have come to the conclusion that not even all data can be trusted, let alone the conclusions. There are many cases of experimenters running dozens of experiments with slightly different conditions before hitting on one that gives them the desired p 0.05. In fMRI experiments, researchers often take more than a year to analyze the results over and over again, changing between different methods of analysis (ROI, threshold, smearing, boxing, statistic, etc.) and then publish the one they (or their supervisor) like best. Or making SPSS return all cross-correlations on questionnaires with over 200 questions and then drawing conclusions from the set of the most significant ones.
So yes, quite a lot of research from "neuro and social psychology" is worthless. I'd say about 90% of it. The problem is just finding out which 10% is valid. And don't start me on replication. Open any journal on experimental psychology and show me articles that replicate a previous experiment *exactly*.
It's not the method that causes the problem, it's the hypothesis. There is some weird, unwarranted reasoning in going from "in a perfect society, gender, race, religion, etc. don't matter" to a null hypothesis that says that the number of male and female characters in a video game should be equal, or any other assumption about distributions. Firsty, the original starting point is not hard (as in gravity or DNA), but subjective. Secondly, the connection between the real world and a video game is superficial at best.
We can also discuss methods, of course, if you feel more comfortable (think representative sample, not being able to manipulate the variables of interest, etc.), because one thing I've learned in my 20 year career in the social sciences is that method is often considered more important than proper reasoning and experimental set-up, usually because that tends to get in the way of obtaining easily publishable results.
A funny tautology, you don't see that often. Witty!
You read the numbers in a weird way. The pages of Secunia say:
- Secunia has issued a total of 193 Secunia advisories in 2003-2009 for Microsoft Windows Server 2003 Enterprise Edition. Currently, 6% (12 out of 193) are marked as unpatched with the most severe being rated Less critical
- Secunia has issued a total of 130 Secunia advisories in 2003-2009 for Apple Macintosh OS X. Currently, 4% (5 out of 130) are marked as unpatched with the most severe being rated Moderately critical
At least on OSX, the most critical vulnerability requires you to download and mount a malcrafted disk image, which most likely can only crash your system, and perhaps expose kernel memory.
Furthermore, it is 4 against 12 unpatched (in favor of OSX), and of course the other Windows systems (XP, Vista, 7) have got a different set of vulnerabilities. And your language and mark-up suggest mild paranoia. You have not been attacked, and the Pravda doesn't deal with Windows viruses. Get treatment, or move to Montana and join the militia.
It's not going to happen in ten years, and I sort of know what I'm talking about (I've been working on neural models of language processing for quite some time now). First, the brain is huge, really huge. The number of "connections" isn't even known, let alone how everything interconnects. Second, even at the level of a single neuron, there are quite a few problems to be solved. If we could figure out the exact structure of a human brain, but emulate it just using integrate-and-fire type neurons, I predict it would fail miserably. Third, brains change. We don't know how we start out, but our brain changes from learning (if you want to be bewildered: development of cat vision is fascinating). For our brain to reach a stage in which it could be subject to a mental disorder takes some 15 years of normal life, i.e. inside a body that functions in a society. So even if someone could build a baby brain in 10 years, and incorporate it into a very human looking body, it would still take 15 years to develop and then might just not have developed a mental disorder.
Fourth, we don't have a clue as to where to look for mental disorders. Of course, if you have a virtual brain, you could tweak parameters, but not only are these parameters at this moment unknown, it would also be unethical (for it would hurt a person, see also other threads), and still this would not offer more insight into the emergent property that mental disorder is as comparing people with these disorders at this moment.
Concluding, this is a seriously flawed project.
I don't like Bonferroni at all, but it does give an indication that the outcome is hardly as significant as the author would like us to believe. But the method of analysis is wrong to begin with. Shouldn't it have been either Kruskall-Wallis (and then use all levels instead of just Excellent vs. non-Excellent) or a chi square? And if he would have included "Very good" at maths, the outcome would have been very different.
So we're just looking at post-hoc interpretation of bad statistics here. Move along, please...
That, and the fact that he fails to publish his method of analysis. What was his hypothesis exactly? He didn't seem to have one in advance, so it seems rather post-hoc. And I don't see where he got his p.01 from, because my intuition tells me 44% vs 56% from a group of 27 respondents cannot be that significant (binomial test). He should have tested the interaction, but the page he refers to has no way to test that, so I think we can take the results with the well-known grain of salt.
We know the bit about lies, damned lies and statistics, of course, but bad statistics...
You are wrong. It is very likely that the user group of this "Mechanical Turk" is highly biased. You *could* say that the conclusion is valid for people belong to this group with a propensity to answer questions of the kind "is a photoshopped nude image of Miley Whatever-her-last-name-is illegal".
As to the discovery of "default", the word exists since the 13th century, according to Merriam-Webster:
Etymology: Middle English defaute, defaulte, from Anglo-French, from defaillir to be lacking, fail, from de- + faillir to fail
Date: 13th century
and the expression "in default of" means "in the absence of". The word "default" was simply chosen by computer scientists to describe what happens in case of absence of specification.
But then again, the whole 13th century probably took place somewhere in the 60's.
Does convoluted writing add credibility to your statement?
Does not knowing the slightest thing about cognitive psychology help you get attention?
Not in the rest of the world, but on /. it gets you to the front page.